# Research and Evidence Back to [[Ribosome Pausing and Cotranslational Folding]] · Post: [[Post Analysis - thegreatest_sv on Ribosome Pausing (2026-10-23)]] · Terms: [[Vocabulary]] All DOIs below were resolved against CrossRef on Oct 24, 2026. Short descriptions are my summaries of titles and abstracts; I did not read every full text. ## 1. The core idea in one paragraph A ribosome reads mRNA three bases (one codon) at a time and adds one amino acid per codon, at roughly 5–6 amino acids per second in mammalian cells ([Ingolia, Lareau & Weissman 2011](https://doi.org/10.1016/j.cell.2011.10.002)). The genetic code is redundant: most amino acids have several **synonymous codons**, and these are decoded at different speeds depending on tRNA supply, wobble pairing and mRNA structure. The growing chain starts folding while it is still being made (**cotranslational folding**). So the *rhythm* of translation, set partly by codon choice, can affect which shape a protein ends up in. Short, **programmed pauses** can help; long **stalls** trigger collisions and quality-control pathways that destroy the mRNA and the protein. ## 2. Timeline of the evidence | Year | Finding | Source | |---|---|---| | 1973 | Anfinsen: a protein's sequence contains the information for its native fold (the baseline the "timing" idea refines) | [Anfinsen, Science](https://doi.org/10.1126/science.181.4096.223) | | 1987 | Proposal that ribosome pausing (slow codons) helps the nascent chain fold in sections | [Purvis et al., J Mol Biol](https://doi.org/10.1016/0022-2836(87)90230-0) | | 1996 | Slow-translating codon clusters fall near domain and structural boundaries | [Thanaraj & Argos, Protein Sci](https://doi.org/10.1002/pro.5560050814) | | 1999 | Synonymous codon substitutions change the specific activity of a protein | [Komar, Lesnik & Reiss, FEBS Lett](https://doi.org/10.1016/s0014-5793(99)01566-5) | | 2002 | Nascent peptides that stall the ribosome on purpose (SecM, TnaC) as regulatory devices | [Nakatogawa & Ito, Cell](https://doi.org/10.1016/s0092-8674(02)00649-9) · [Gong & Yanofsky, Science](https://doi.org/10.1126/science.1073997) | | 2006 | Stalled ribosomes trigger mRNA cleavage (no-go decay) | [Doma & Parker, Nature](https://doi.org/10.1038/nature04530) | | 2007 | A "silent" MDR1 polymorphism changes P-glycoprotein substrate specificity | [Kimchi-Sarfaty et al., Science](https://doi.org/10.1126/science.1135308) | | 2008 | "Codon harmonization": keep the native slow/fast pattern when moving genes between hosts | [Angov et al., PLoS ONE](https://doi.org/10.1371/journal.pone.0002189) | | 2009 | Speeding translation through slow clusters reduces correct folding (*E. coli*) | [Zhang, Hubalewska & Ignatova, NSMB](https://doi.org/10.1038/nsmb.1554) | | 2009 | **Ribosome profiling** maps every ribosome position at codon resolution | [Ingolia et al., Science](https://doi.org/10.1126/science.1168978) | | 2011 | Review: synonymous mutations in human disease and therapeutics | [Sauna & Kimchi-Sarfaty, Nat Rev Genet](https://doi.org/10.1038/nrg3051) | | 2013 | Codon optimality patterns tied to structure are conserved across yeast species | [Pechmann & Frydman, NSMB](https://doi.org/10.1038/nsmb.2466) | | 2013 | Proteotoxic stress causes elongation pausing | [Liu, Han & Qian, Mol Cell](https://doi.org/10.1016/j.molcel.2012.12.001) | | 2013 | Codon usage tunes FRQ structure and circadian clock function (Neurospora) | [Zhou et al., Nature](https://doi.org/10.1038/nature11833) | | 2014 | Critical analysis of codon optimization in human therapeutics | [Mauro & Chappell, Trends Mol Med](https://doi.org/10.1016/j.molmed.2014.09.003) | | 2015 | Review: synonymous codons and cotranslational folding | [Chaney & Clark, Annu Rev Biophys](https://doi.org/10.1146/annurev-biophys-060414-034333) | | 2015 | Cotranslational folding of CFTR and translation speed | [Kim et al., Science](https://doi.org/10.1126/science.aaa3974) | | 2015 | Codon usage influences cotranslational folding in vivo (Neurospora) | [Yu et al., Mol Cell](https://doi.org/10.1016/j.molcel.2015.07.018) | | 2016 | Synonymous codons direct the same protein to different conformations | [Buhr et al., Mol Cell](https://doi.org/10.1016/j.molcel.2016.01.008) | | 2017 | Ribosome collisions are the trigger for no-go decay | [Simms et al., Mol Cell](https://doi.org/10.1016/j.molcel.2017.08.019) | | 2018 | ZNF598 senses collided ribosomes | [Juszkiewicz et al., Mol Cell](https://doi.org/10.1016/j.molcel.2018.08.037) | | 2019 | Optical tweezers: translation suppresses misfolding of the nascent chain | [Alexander et al., Nat Commun](https://doi.org/10.1038/s41467-019-10647-6) | | 2019 | Codon-optimized coagulation factor IX differs in conformation from wild type | [Alexaki et al., Sci Rep](https://doi.org/10.1038/s41598-019-51984-2) | | 2019 | Review: "Ribosome pausing, a dangerous necessity for co-translational events" | [Collart & Weiss, NAR](https://doi.org/10.1093/nar/gkz763) | | 2019 | Review: elongation kinetics and folding | [Stein & Frydman, JBC](https://doi.org/10.1074/jbc.rev118.002814) | | 2020 | Synonymous codon substitutions perturb cotranslational folding in vivo | [Walsh et al., PNAS](https://doi.org/10.1073/pnas.1907126117) | | 2021 | Disome-seq: widespread collisions that promote cotranslational folding | [Zhao et al., Genome Biol](https://doi.org/10.1186/s13059-020-02256-0) | | 2024 | Machine learning on RiboSeq data links decoding speed to folding | [Aguilar Rangel, Stein & Frydman, Sci Adv](https://doi.org/10.1126/sciadv.ado0738) | | 2026 (preprint) | Codon-supply metrics show negligible independent coupling to structure; downstream mRNA structure gives a small signal ⚠️ not peer reviewed | [Hassan, Jabeen & Kiani, bioRxiv](https://doi.org/10.64898/2026.09.27.753847) | ## 3. What the post's TIME / MARK / SORT / FOLD steps map to | Post step | Real method (pre-existing) | |---|---| | TIME: "follow protein production second by second" | Pulse-chase, real-time translation assays, single-molecule FRET | | MARK: "record exactly where the ribosome stops" | Ribosome profiling / Ribo-seq ([Ingolia 2009](https://doi.org/10.1126/science.1168978)); disome-seq for collisions | | SORT: "separate random slowdowns from recurring pauses" | Pause-site calling across replicates; ML models of decoding speed ([Aguilar Rangel 2024](https://doi.org/10.1126/sciadv.ado0738)) | | FOLD: "check what the growing protein is doing" | Optical tweezers ([Alexander 2019](https://doi.org/10.1038/s41467-019-10647-6)), force-profile arrest-peptide assays, cryo-EM of ribosome–nascent chain complexes | None of these need, or are done by, a general-purpose LLM. Their outputs can be analyzed with code an LLM writes, but no such study with Claude + Jev was found. ## 4. What Anthropic actually published on biology (2026) - **[How Claude is uplifting biomolecular modeling](https://www.anthropic.com/research/claude-uplifts-biomolecular-modeling)** (Sep 17, 2026). Claude optimized 30+ open-source biomolecular models (about 4× faster on average), wrote "FlashPairformer" kernels, and built a low-memory "Big" mode that predicted structures over 10,000 tokens on one GPU node: respiratory complex I, the TRiC chaperonin, a proteasome and **a bacterial 70S ribosome**. Models named: Mythos 5.1, Mythos 5, Opus 5. **No pausing, codon or folding-kinetics result and no mention of Jev.** - **[Autonomous de novo protein binder design with Claude](https://www.anthropic.com/research)** (Aug 18, 2026; linked to Anthropic's research index): 354 of 1,320 designs bound their targets (27%). - "Claude Opus 5.5", named in the video, does not appear in these Anthropic pages ⚠️ (I did not check whether such a model has been released). ## 5. What Jev is - **TypeSafe**, ["Introducing System One Models & Jev"](https://typesafe.ai/blog/introducing-system-one-models-and-jev) (Sep 15, 2026). Jev is a "System One" model: you send a prompt and a typed schema, and it returns a **decision** (choice among options, a numeric score, or "noul", meaning none of the above) with calibrated probabilities. It does **not generate text**. Docs: docs.typesafe.ai/api/system-one. Also listed on OpenRouter (as "Jev 1.14" when checked). - The company says it emerged from stealth with $40M (Business Wire, Sep 2026) ⚠️ (via a community "awesome-jev" list; press release not opened). Founder Diogo Almeida is described as ex-OpenAI; a Diogo Almeida is an author of InstructGPT ([Ouyang et al. 2022](https://arxiv.org/abs/2203.02155)) ⚠️ (same person is likely, not confirmed). - The author's own article says TypeSafe paused new signups on Sep 22, 2026. - **Bottom line:** a fast classifier for routing and triage. It cannot analyze ribosome-profiling data or protein folding by itself. ## 6. Sanity checks on the video's numbers ⚠️ (my arithmetic) | Panel | Video value | Reality check | |---|---|---| | ATP SYNTHESIS (per cell) | 36–37 mol/s | Whole body ≈ 0.0016 mol/s ([Törnroth-Horsefield & Neutze 2008](https://doi.org/10.1073/pnas.0810654106)) ÷ ~3×10¹³ cells ([Sender et al. 2016](https://doi.org/10.1371/journal.pbio.1002533)) ≈ 5×10⁻¹⁷ mol/s. Off by ~10¹⁸. | | PROTEIN SYNTHESIS | ~11,600–12,800 aa/s | ~5.6 aa/s per ribosome × millions of ribosomes per mammalian cell → millions of aa/s. Likely ~1000× low. | | Ribosome size | 0.022 µm | ~20–30 nm is the right order of magnitude ✅ | | Lysosome pH | 4.6 | Within the usual ~4.5–5 range ✅ | | Readouts between frames | cytosol 52.5% ↔ 55.4%, protein synthesis 12,587 ↔ 12,770 aa/s | Values animate continuously, consistent with a motion graphic, not a measurement |